MATLAB: How to create a good mask of the algae without reading into too much noise

algaedigital image processingimage processingImage Processing Toolboximage segmentation

The goal is to produce a (binary) mask on the algae and I attempted following steps:

– turn image gray with default function rgb2gray()

– image smoothing by gaussian filtering with sigma 0.5

– obtain image gradient with ingredient() using 'CentralDifference'

– Sobel Edge detection

– Canny Edge detection 2 times

– image dilation with square shape structural elements size 2 pixels and fill in

– image erosion with diamond shape size 3 pixels

The result is not good. The main problem is that my program has a hard time figuring out the ambiguous contour of the original image. It also reads into too much noise (I suspect it will get better if I increase the sigma of gaussian filter)

Can someone help me with this? I have some other 500 images that I need to create masks for. They are all of the same color tone (gray green type)

Best Answer

The problem is that the algae virtually disappears at points. So this is actually one situation where an edge filter might actually help. So try some. I see you did and weren't happy with the results. Well those were easy ones to try since they're built in, but perhaps not the best. You might try a DOG filter (Difference of Gaussians). Or try a Frangi or Hessian filter - there are File Exchange entries for those. If you still get a tube with choppy edges, then you'll have to try to make them straighter. In that case either try activecontour() (demo attached), or maybe Hough() or houghlines(). If the algae tubules are always straight lines, then you might be able, and need to, use RANSAC. You could avoid this if you could somehow stain the algae, or use fluorescence photography if fluoresces.

The easy part is to get rid of the small noise blobs once you have the binary image. Just use bwareaopen() or bwareafilt().

Related Solutions

MATLAB: Image Prcoessing – Automatic Region of Interest Extraction – Help!!

Ok here is what I could do in 5 mins with my above suggestions and a little more:

1. Get Blue channel with the green line on the image:

2. Convert to double, take square of the image, erode with a disk strel until satisfied:

 XPX = double(blue);
 XPX = XPX.^2;
 se = strel('disk',3,0)'
 XPX = imerode(XPX,se);

3. Threshold and find biggest connected component inside the green boundary, then fill holes:

4. Activecontour with the thresholded image as a mask, then imopen with a giant disk to round:

 bw3 = activecontour(blue,BW2,300);
 se3 = strel('disk',100,0);
 BWXP3 = imopen(bw3,se3);

5. ROI Overlayed:

I am sure you can get better results with more fine tuning or maybe resizing to a smaller image, however this process makes some specific assumptions about the nature of your data, which may not be applicable for the rest of your images as Image Analyst mentioned. He would also have more experience with this stuff, I only do computer vision in a narrow field of study.

MATLAB: Making a histogram and limiting gaussian filter to a boundary

You can either take the histogram of the whole image and suppress the bin at gray level 0:

[pixelCounts, grayLevels] = imhist(grayImage);\
pixelCounts(1) = 0; % Zero out the black bin.

or you can have a mask and take the histogram of pixels only within the mask:

[pixelCounts, grayLevels] = imhist(grayImage(mask));

To make filter work only within mask you have to decide what you want to do at the boundary. Do you want it to consider pixels outside the boundary and just mask off the result?

filteredImage = imfilter(grayImage, kernel); % Filter whole image

filteredImage(mask) = 0; % Mask off result.

Or do you want to consider pixels outside as zero?

grayImage(~mask) = 0; % Zero outside mask.
filteredImage = imfilter(grayImage, kernel); % Filter whole image

Neither is right or wrong - it just depends on how you want to handle it.

Best Answer

Related Solutions

MATLAB: Image Prcoessing – Automatic Region of Interest Extraction – Help!!

MATLAB: Making a histogram and limiting gaussian filter to a boundary

Related Question